Few-shot learning using pre-training and shots, enriched by pre-trained samples
Machine Learning
2020-09-22 v1 Computer Vision and Pattern Recognition
Abstract
We use the EMNIST dataset of handwritten digits to test a simple approach for few-shot learning. A fully connected neural network is pre-trained with a subset of the 10 digits and used for few-shot learning with untrained digits. Two basic ideas are introduced: during few-shot learning the learning of the first layer is disabled, and for every shot a previously unknown digit is used together with four previously trained digits for the gradient descend, until a predefined threshold condition is fulfilled. This way we reach about 90% accuracy after 10 shots.
Keywords
Cite
@article{arxiv.2009.09172,
title = {Few-shot learning using pre-training and shots, enriched by pre-trained samples},
author = {Detlef Schmicker},
journal= {arXiv preprint arXiv:2009.09172},
year = {2020}
}